Complex Thermal Expansion Properties in aMolecular Honeycomb Lattice

Jonathan J. Loughreya,b, Tim P. Comync, David C. Apperleyd,Marc A. Littlea,e and Malcolm A. Halcrow*a

aSchool of Chemistry, University of Leeds, Woodhouse Lane, Leeds, UK LS2 9JT.E-mail: m.a.halcrow@leeds.ac.uk

bCurrent address: Department of Chemistry and Biochemistry, University of Arizona,P.O. Box 210041, 1306 East University Blvd., Tucson, AZ 85721-0041, USA

cInstitute for Materials Research, School of Process, Environmental and Materials Engineering,University of Leeds, Clarendon Road, Leeds, UK LS2 9JT.

dEPSRC Solid State NMR Service, Department of Chemistry, University of Durham, South Road,Durham, UK DH1 3LE.

eCurrent address: Department of Chemistry, University of Liverpool, Crown Street, Liverpool,UK L69 7ZD.

Electronic Supplementary Information

Experimental procedures and characterisation data.

Table S1 Experimental data for the single crystal structure determinations of 1.

Figure S1 Contents of the asymmetric unit of 1 at 300 K and 110 K.

Table S2 Selected interatomic distances and angles in the crystal structures of 1.

Table S3 Hydrogen bond parameters for the crystal structures in this work.

Table S4 Variable temperature unit cell parameters for 1 from single crystal and powder diffraction data.

Figure S2 An alternative set of single crystal unit cell data from 1, measured using a different crystal.

Figure S3 Observed and simulated X-ray powder diffraction data from the different phases of 1.

Figure S4 Changes to the X-ray powder diffraction pattern of 1 during in situ dehydration.

Figure S5 Thermogravimetric analyses (TGAs) of 1and 2.

Figure S6 Solid state 11B NMR spectra of 1, at 298, 238 and 168 K.

Figure S7 Solid state 19F NMR spectra of 1, at 298, 238 and 168 K.

Figure S8 Variable temperature solid state 11B NMR linewidths.

References

Electronic Supplementary Material (ESI) for Chemical Communications.This journal is © The Royal Society of Chemistry 2014

Synthetic Procedures and Characterisation Data for the Compounds in this Work.

3{5}-(Pyrazinyl)-1H-pyrazole (L) was prepared by the literature procedure,[1] while all other reagentswere purchased commercially and used as supplied. The complex [FeL3][BF4]2·xH2O (1) was preparedby mixing aqueous solutions of L (0.25 g, 1.7 mmol) and Fe[BF4]2·6H2O salt (0.19 g, 0.6 mmol). Slowevaporation of the resultant deep red solution to dryness afforded clusters of brown needle-shaped crystals thatwere collected, dried and analysed without further purification. Found C, 35.2; H, 3.10; N, 23.2 %. Calcd.for C21H18B2F8FeN12·3H2O C, 35.0; H, 3.35; N, 23.3 %.

Elemental microanalyses were performed by the University of Leeds School of Chemistrymicroanalytical service. Thermogravimetric analyses employed a TA Instruments TGA 2050 analyser.X-ray powder diffraction was collected using a PANalytical X’Pert MPD diffractometer (Cu K1+2),with an in situ liquid N2 cold stage (Anton Paar TTK450). Data were collected between 2 = 5 and50°, with a step size of 0.033°; the total time per scan was 20 mins. Rietveld refinements wereprepared using PANalytical X’Pert Highscore Plus, using the low temperature single crystalrefinement as an initial model.

Solid-state spectra were recorded at 96.29 (11B) or 282.40 (19F) MHz using a Varian UnityInova spectrometer and a 4 mm (rotor o.d.) magic-angle spinning probe. They were obtained usingdirect excitation with a 2 s (11B) or 5 s (19F) recycle delay at a sample spin-rate of 10 kHz. The numberof repetitions was 32 (11B) and 8 (19F). Heating/cooling rates were approximately 5° per minute.Chemical shifts are referenced to an external sample of BF3∙OEt2 or CFCl3.

Experimental Procedures for the Crystal Structure Determinations

The full diffraction datasets were collected using a Bruker X8 Apex II diffractometer, with graphite-monochromated Mo-K radiation ( = 0.71073 Å) generated by a rotating anode. The diffractometerwas fitted with an Oxford Cryostream low temperature device. The structures were solved by directmethods (SHELXS97[2]), and developed by cycles of full least-squares refinement on F2 and differenceFourier syntheses (SHELXL97[2]). All crystallographic Figures were produced using XSEED,[3] whichincorporates POVRAY.[4] Experimental data for the crystal structures are listed in Table S1.

The variable temperature unit cell data reported in the manuscript and the ESI were measuredwith an Agilent Supernova dual-source diffractometer, again using monochromated Mo-Kα radiation.The crystal was cooled or warmed between measurements at 2 Kmin−1 under computer control, andpoised for 10 mins at each temperature before the measurement was taken.

Crystallographic refinement of [FeL3][BF4]2·xH2O (1; x ≈ 3).

The same crystal was used for data collections before and after rapid cooling to 110 K, while adifferent crystal was used to obtain data at 300 K before, and after, slow cooling to 150 K andrewarming to room temperature. The refinement details were the same in each case. In the slowcooling experiment, cooling below 250 K resulted in twinning of the crystal which was reversed uponre-warming.

The asymmetric unit contains one-third of a complex cation and one-third of a BF4– anion, all

lying on a crystallographic three-fold axis; and, a very badly disordered region, which lies within achannel running parallel to the c-vector of the unit cell. These channels contain another one-thirdequivalent BF4

– ion (which is required for electroneutrality), and approximately one mole equivalent ofwater per asymmetric unit. A SQUEEZE[5] analysis of all four datasets showed that the channelscontain 61 between and 79 electrons per formula unit, with the highest value being observed for thelow temperature dataset. Despite the scatter, those values agree reasonably with one BF4

– and threewater molecules (71 electrons in total) as predicted by elemental microanalysis.

The initial datasets, rather than the SQUEEZED ones, were used for the final refinements. For theslow-cooled crystal, at both temperatures the strongest Fourier peaks in the disordered region wererefined as two partial BF4

– sites, whose occupancies were refined against the average isotropic Ueq

value of the ordered BF4– ion B(13)-F(15) at that that temperature. At 300 K the occupancies of these

two sites refined to 0.10 and 0.09, which corresponds to 58 % of the total BF4– content in the

disordered channels. At 110 K the same anion site occupancies refined to 0.14 and 0.12, or 78 % of thetotal disordered BF4

– content. The refined restraints B–F = 1.40(2) and F...F = 2.29(2) Å were appliedto this residue in both refinements. For the slow-cooled and rewarmed crystal, only one partial BF4

–

site could be resolved for each refinement with an occupancy of 0.10, or 30 % of the total anioncontent in the pores. The distance restraints in this case refined to lower values, of B–F = 1.35(2) andF...F = 2.20(2) Å.

The remaining contents of the asymmetric unit were not included in the final models, but areaccounted for in the density and F000 calculations. All crystallographically ordered non-H atoms wererefined anisotropically, and all H atoms were placed in calculated positions and refined using a ridingmodel. CCDC 992761−992764.

Table S1 Experimental data for the single crystal structure determinations of 1 (C21H24B2F8FeN12O3, Mr 721.99, Trigonal, space group P31c, Z = 2)

Before rapid cooling(T = 300 K)

After rapid cooling(T = 110 K)

Before slow cooling:(T = 300 K)

After slow cooling to 150 Kand rewarming (T = 300 K)

a (Å) 12.6729(5) 12.6653(9) 12.6778(7) 12.660(2)c (Å) 10.9338(6) 11.0258(12) 10.9308(8) 10.9359(15)V (Å3) 1520.74(12) 1531.7(2) 1521.49(16) 1517.9(4) (Mo-K) (mm–1) 0.592 0.588 0.592 0.592Measured reflections 11971 13308 21264 12294Independent reflections 2366 1850 2090 1709Rint 0.040 0.046 0.041 0.033R1 [I > 2(I)]a 0.038 0.050 0.043 0.042wR2 (all data)b 0.107 0.144 0.122 0.128Goodness of fit 1.048 1.117 1.162 1.112Flack parameter 0.010(18) 0.01(2) 0.04(2) 0.03(3)

aR = [Fo – Fc] / FobwR = [w(Fo

2 – Fc2) / wFo

4]1/2

Fig S1 Contents of the asymmetric unit of 1 at 300 K (left) and 110 K (right), taken from the “rapid cooling” experiment. Displacement ellipsoids are atthe 50 % probability level, and C-bound H atoms have been omitted for clarity. Colour code: C, grey; H, pale grey; B, pink; F, yellow; Fe, green; N,blue.

Atoms Fe(1), B(13) and F(14) all lie on the crystallographic C3 axis ⅔, ⅓, z. Symmetry codes: (i) 1–y, x–y, z; (ii) 1–x+y, 1–x, z; (iii) 1+x–y, 1–y, –½+z;(iv) x–y, 1–y, ½+z.

Table S2 Selected interatomic distances and angles in the crystal structures of 1 (Å, °). See Fig. S1 for the atomnumbering scheme employed. Symmetry code: (i) 1–y, x–y, z.

Before slowcooling(T = 300 K)

After slowcooling(T = 110 K)

Before fastcooling(T = 300 K)

After fast cooling to150 K and rewarming(T = 300 K)

Fe(1)–N(2) 1.985(2) 1.984(3) 1.988(2) 1.982(3)Fe(1)–N(9) 1.967(2) 1.963(3) 1.967(3) 1.969(3)

N(2)–Fe(1)–N(9) 80.25(8) 80.47(11) 80.31(10) 80.26(13)N(2)–Fe(1)–N(2i) 94.80(7) 94.33(10) 94.67(9) 94.54(12)N(2)–Fe(1)–N(9i) 88.87(9) 90.11(12) 88.83(11) 89.22(14)N(9)–Fe(1)–N(2i) 174.09(10) 173.42(13) 174.12(12) 173.82(14)N(9)–Fe(1)–N(9i) 96.33(8) 95.42(10) 96.43(10) 96.25(12)

Table S3 Hydrogen bond parameters for the crystal structures in this work (Å, °). See Figs. S1 and S2 for theatom numbering schemes employed. Symmetry codes: (iii) 1+x–y, 1–y, –½+z; (vii) –1+y, x, ½+z.

D−H H…A D…A D−H…A 1, before slow cooling (T = 300 K)C(4)–H(4)…F(17A)/F(17B) 0.93 2.55/2.40 3.383(19)/3.30(3) 149.5/163.1N(10)–H(10)…N(5iii) 0.86 2.02 2.826(3) 156.7

1, after slow cooling (T = 110 K)C(4)–H(4)…F(17A)/F(17B) 0.95 2.49/2.45 3.354(14)/3.35(2) 150.8/157.7N(10)–H(10)…N(5iii) 0.88 1.98 2.810(4) 156.4

1, before fast cooling (T = 300 K)C(4)–H(4)…F(17) 0.93 2.44 3.29(2) 152.3N(10)–H(10)…N(5iii) 0.86 2.02 2.824(4) 155.8

1, after fast cooling and rewarming (T = 300 K)C(4)–H(4)…F(17) 0.93 2.45 3.31(2) 152.8N(10)–H(10)…N(5iii) 0.86 2.01 2.822(5) 156.1

Table S4 Unit cell parameters for 1 from single crystal diffraction data, and from Rietveld refinement of X-raypowder diffraction data. These data are plotted in Fig. 3 of the main paper.Single crystal PowderT (K) a (Å) c (Å) V (Å3) T (K) a (Å) c (Å) V (Å3)290 12.6633(6) 10.9157(5) 1515.93(12) 298 12.690(2) 10.914(3) 1522.1(7)280 12.6573(6) 10.9227(5) 1515.47(13)270 12.6474(6) 10.9242(5) 1513.34(12) 273 12.666(2) 10.922(3) 1517.4(7)260 12.6405(5) 10.9349(4) 1513.13(10)250 12.6362(5) 10.9325(5) 1511.77(11) 248 12.660(2) 10.931(3) 1517.4(7)240 12.6296(6) 10.9338(5) 1510.36(12)230 12.6288(6) 10.9380(5) 1510.76(13)220 12.6398(10) 10.9440(8) 1514.2(2) 223 12.655(2) 10.941(3) 1517.5(7)210 12.6438(13) 10.9465(10) 1515.5(3)200 12.6446(16) 10.9477(12) 1515.9(3) 198 12.652(3) 10.958(4) 1519.0(7)190 12.6397(18) 10.9458(13) 1514.4(4)180 12.647(2) 10.9492(13) 1516.8(4)170 12.637(2) 10.9461(15) 1513.8(5) 173 12.646(3) 10.981(5) 1520.8(7)160 12.643(3) 10.9530(17) 1516.1(6)150 12.663(3) 10.9683(19) 1523.1(6) 148 12.634(4) 11.000(5) 1520.6(8)140 12.634(4) 10.9517(19) 1513.8(7)130 12.630(3) 10.9584(19) 1513.9(6)120 12.623(3) 10.9758(18) 1516.2(6) 123 12.623(3) 11.005(5) 1518.7(7)110 12.624(3) 10.9778(17) 1515.2(5)100 12.610(2) 10.9832(15) 1512.5(4)110 12.640(2) 10.9937(16) 1521.0(5)120 12.632(3) 10.9791(17) 1517.2(5)130 12.621(3) 10.9621(17) 1512.1(6)140 12.631(3) 10.9621(18) 1514.6(6)150 12.642(3) 10.9621(17) 1517.3(6)160 12.649(3) 10.9625(17) 1519.0(5)170 12.642(3) 10.9537(16) 1516.0(5)180 12.646(2) 10.9504(14) 1516.5(4)190 12.6410(18) 10.9451(13) 1514.7(4)200 12.6442(15) 10.9474(11) 1515.8(3)210 12.6369(13) 10.9420(10) 1513.3(3)220 12.6341(10) 10.9423(8) 1512.6(2)230 12.6298(7) 10.9386(6) 1511.08(16)240 12.6337(6) 10.9444(5) 1512.80(12)250 12.6493(6) 10.9429(5) 1516.32(12)260 12.6499(5) 10.9363(5) 1515.56(11)270 12.6448(4) 10.9290(4) 1513.34(9)280 12.6495(5) 10.9222(4) 1513.50(10)290 12.6476(5) 10.9166(4) 1512.28(10)

Fig S2 An alternative set of single crystal unit cell data from 1, measured under the same conditions as the data inFig. 3 (main article) and Table S4, but using a different crystal.

These data follow the trends in the manuscript between 300→220 K, but are too noisy at lower temperatures to interpret meaningfully. Five crystals were measured in this way during the course of this study, using twodiffractometers. All the crystals showed increased errors and noise in their cell parameters below 220±20 K. Thedata in Fig. 3 and Table S4 also have that property, but to a lesser extent than the other crystals we examined.

Fig S3 Observed and simulated X-ray powder diffraction data from the different phases of 1. Data were measuredin variable slit mode, and are uncorrected.

There are diffraction peaks at 2 = 24 and 31°, and at 2 > 35°, which do not have counterparts in the simulation.These may originate from the pore contents which are not resolved in the single crystal refinements and so couldnot be included in the simulation.

There are no changes to the form of these powder patterns upon cooling, apart from a gradual peak broadeningbelow 220 K which is reversed when the sample is rewarmed. This reduction in crystallinity is consistent with theincreased errors on the single crystal unit cell parameters at lower temperatures (Figs. 3 and S2).

Fig S4 Changes to the X-ray powder diffraction pattern of 1 during in situ dehydration at 10–5 torr and 298 K.Data were measured in variable slit mode, and are uncorrected.

The dehydration reaction is essentially complete after 4 hrs. The dehydrated material reconverts fully to theinitial phase upon exposure to air, within a period of minutes.

Fig S5 Thermogravimetric analysis (TGA) of 1. The weight losses equivalent to two and three mole equivalentsof water are marked.

The curve has an inflection near 440 K, at a weight loss equivalent to two mole equivalents of lattice water. Theweight loss equivalent to a third mole of water occurs near 500 K, but is less well defined.

Fig S6 Solid state 11B NMR spectra of 1, at 298 (green), 238 (red) and 168 K (black).

Fig S7 Solid state 19F NMR spectra of 1, at 298 (green), 238 (red) and 168 K (black).

For each nucleus we assign the broader resonance to the more static, crystallographically ordered framework BF4−

ion; and, the narrower resonance to the (dynamically) disordered in-pore BF4− environment.

The weak feature at −149 ppm occurs consistently between samples, and may arise from partial hydrolysis of the in-pore BF4

− ions to BF3(OH)−.[6]

Fig. S8 11B NMR linewidths for the in-pore (circles) and framework (squares) BF4− ions. Data were measured

using the following temperature ramps: slow cool (black); slow rewarm (grey); and rapid quench then rewarm(white).

These changes with temperature are smaller than for the 19F linewidths plotted in the main article, particularly forthe in-pore environment. Although the shape of the in-pore anion curve appears to follow the variable temperatureunit cell dependence of 1, the variation is barely outside the error of the experiment.

References

[1] K. L. V. Mann, J. C. Jeffrey, J. A. McCleverty and M. D. Ward, Polyhedron, 1999, 18 721.[2] G. M. Sheldrick, Acta Cryst., Sect. A, 2008, 64, 112.[3] L. J. Barbour, J. Supramol. Chem., 2001, 1, 189.[4] POVRAY v. 3.5, Persistence of Vision Raytracer Pty. Ltd., Williamstown, Victoria, Australia,

2002.[5] A. L. Spek, J. Appl. Cryst., 2003, 36, 7.[6] R. J. Smithson, C. A. Kilner, A. R. Brough and M. A. Halcrow, Polyhedron, 2003, 22, 725.